The if-then-else monster. Bloated functions that take dozens of kwargs. When you read the code you can't even tell what runs because the cross-product of all the configurations is beyond human comprehension. Majority of the paths are deprecated, unsupported, or unadvisable.

Getting employees to work hard and deliver really isn't a matter of mandating work-from-office and long hours. It's a matter of incentives and ownership. People do their best when they work on interesting problems, in a self-directed manner, and get rewarded for success. This

The academic year has officially begun! During Mila’s welcome week, our new cohort of students had the opportunity to explore our facilities, engage in a poster session, participate in information sessions, and wrap up the week with a social event. Welcome to all!

Heavy Is The Crown.

Another day, another TorchRL application!

For friends of open source: imo the highest leverage thing you can do is help construct a high diversity of RL environments that help elicit LLM cognitive strategies. To build a gym of sorts. This is a highly parallelizable task, which favors a large community of collaborators.

I am hiring another postdoc for my lab. Consider applying if you have #foundationalmodels, #robotics, or #reinforcementlearning skills. You will help create generalist real-world agents (robots) with a team of 20 working on these problems and overly competitive go-karting.

I wrote this last year, and it doesn't hurt to share it again. A phone-formatted ELBO TL;DR.

Can some ML people tell me? Who is Adam?

My lab with REAL - Robotics and Embodied AI Lab was honoured to present our recent research and explain how machine learning enables robotics that will drive innovation in Canada. It was a pleasure getting into the details!

A.I. is the key to unlocking productivity, higher paying jobs, and more prosperity.   I met with the team at the Mila A.I. Institute yesterday to discuss how Canada can be at the forefront of emerging technology that will make Canada’s economy stronger and more competitive.

VLAs offer an avenue for generalist robot policies; however, naively following the action predictions leads to brittle or unsafe behaviours. We introduce VLAPS, which integrates model-based search with pre-trained VLA policies to improve performance without additional training.